"Back To The Future" using Generative Design & Investment Casting

[SteveCox3D 87]

How can the very latest, cutting-edge design software combine with a 5,000 year old manufacturing technique to deliver outstanding weight reduction opportunities?

 

Designing for light-weight parts is becoming more important, and I’m a firm believer in the need to produce lighter weight, less over-engineered parts for the future. This is for sustainability reasons because we need to be using less raw materials and, in things like transportation, it impacts upon the energy usage of the product during it’s service life. Lighter products mean less fuel to move them around, which can make our fossil fuel reserves go further, or make more efficient use of the renewable energies that we’re now beginning to adopt.

 

Generative Design (GD) is the very latest design software released by Autodesk and is now included in Fusion 360, which is at the heart of their "Future of Making Things" strategy for Design and Manufacturing. It changes the way we design things and can deliver very efficient designs that deliver structural performance with optimised use of material.

 

 

The aerospace industry is expected to be one of the early adopters of this technology because in that industry the cost and environmental savings from improved fuel efficiency carry the greatest rewards. Also, I see interest from the automotive industry for the same fuel efficiency reasons, but in the long term the drive for lighter weight parts could benefit many industries, even those outside of transportation.  Another example of the benefits of lighter weight alongside reduced material usage is that shipping costs for parts reduce as their weight reduces, which can therefore also deliver cost efficiencies.

 

GD is targeted initially at metal parts where the biggest opportunity for light-weighting exists. The complex forms it generates though often means that parts conceived in this way cannot be made with conventional manufacturing routes. They therefore need to use Additive Manufacturing (AM) techniques to produce them.

 

 

The route of using high energy, laser-based AM to do this comes with associated high costs because of the specialised set-up knowledge required together with expensive processing, and post processing, to deliver a quality-assured part. This project explores the possibility of a more cost-effective route to a metal GD part which, even though at this stage may be just used for a small quantity of evaluation prototypes, can act as an enabler for understanding the potential that GD has to offer.

 

This is the baseline design for this project. It is an aluminium bracket design similar to those used in aerospace applications to mount control surfaces, and in this form has not been optimised for weight. This design would weigh 383 grams in the intended material, aluminium A356.

 

 

After processing this through Generative Design in Fusion 360 it’s time to review and evaluate the many alternative design options presented and decide upon the design that is considered the most appropriate taking into the other factors that have an influence on design selection such as manufacturability, aesthetics etc.

 

 

This was the design option chosen for this part and Fusion 360 was used to create the final version of the model.

 

 

The bio-mimicry that’s evident in most of the designs created by GD is interesting to see, in this case the design of the part can be seen as essentially a swept I-beam (which engineers, especially those in construction, are taught is a strong section), but with tendon-like attachments back to the mounting points to carry the tensile loading that’s created by the applied loading conditions

 

What GD does is to turn the standard design workflow that we’re familiar with on it’s head. Traditionally we design a part and then stress test it virtually to determine if it fulfils the required structural performance. Any failures seen during this process require an iterative loop back to the design to correct them.

 

 

With GD the stress analysis is a core part of the design synthesis, and happens as the part design iterates, which means that the output at the end should meet the requirements of the intended loading requirements. The software is searching for an optimal solution where the stress is ideally evenly distributed across the part as can be seen above.

 

To prove that everything is good with the finalised design this part has then been virtually tested again in Fusion 360 to confirm that the original loading requirements are still met

 

 

So we've created our lightweight part design, and maybe now we need to produce that in aluminium A356 to do some physical testing, but don’t want the expense of using a metal AM process. What follows is a way of achieving this where FDM 3D printing can play a role as an “enabler” to help create the final parts in conjunction with a very old (if not ancient) manufacturing technique called investment casting. This technique is 5,000 years old according to Wikipedia.

 

The company involved with casting this project is Sylatech who have been using Ultimaker 3D printers as part of their process for investment casting of prototype parts

 

Sylatech took the .stl file of this model and used it to create a 3D print of the part on an Ultimaker 3 in PLA. This PLA part was then used as the pattern in the investment casting process where it is submerged in plaster under vacuum conditions to ensure that all air is excluded from the mould and creates an accurate reproduction of the surfaces of the part. The picture below shows a display box which demonstrates the set up of the 3D printed parts partially encased in plaster. 

 

 

Once the plaster has hardened the casting box is put into a furnace at very high temperature in order to burn out the PLA, leaving behind a cavity into which molten aluminium can be cast.

After solidification of the metal, and cooling of the mould, the plaster is broken away from the parts, and then they can be quickly and easily removed from the material feed gate resulting in these aluminium A356 versions of the PLA original.

 

 

The final part weighs 122 grams which is a weight saving of 68% over the original baseline part, which shows the potential that GD has to make significant reductions in weight and material usage. Using this method we now we have an excellent quality physical part made very quickly in the final intended material in order to commence some physical testing.This is a different route to get to that physical test part in metal at a fraction of the cost of having it metal additively manufactured. 

 

It also shows how a brand new, cutting edge piece of software that only became available in May 2018 can combine with FDM 3D printing (which many people still see as a new technology even though it’s been around for over 20 years) and a 5,000 year old manufacturing technique to deliver potentially huge benefits in weight and material usage.

 

Using the investment casting route in this case study is why I chose the title for this article, and shows that we can effectively go “Back To (Deliver) The Future”.

 

Do you see the need for lighter weight parts in what you do, and can you see the potential benefits of using Generative Design and this method of producing metal parts? 

 

I'd welcome comments, suggestions, and discussion about any aspects of the above article, the next steps that I'm looking at are how this process could scale up to batch production of the parts using 3D printing techniques that could support low volume production quantities

 

 

 

 

Edited October 11, 2018 by SteveCox3D
Added picture

[JohnInOttawa 104]

That's a great article Steve.  Thank you.

 

The benefits of the weight savings are incredible and the part looks nothing like I would imagine, so under the heading of 'thinking outside the box', this approach is game changing.  Not only is there savings in building the part itself, but there is huge potential to save fuel, reduce emissions and, conversely, make parts much, much stronger than they are today for an equivalent weight.   And guys like me are very partial to strong parts ?

 

What follows is largely beyond the scope of the original discussion, but I see it as an inevitable crossroads as we refine these kinds or parts.

 

One of the wild cards we seem to struggle with still is how to handle abuse cases in storage, handling, installation and in-service damage. 

 

We have an industry that grew up on over-built, over-engineered parts.  An unintended benefit of some (not all) of these components is that they could withstand (often undocumented) conditions outside of the original design scope without exhibiting damage.   (I need to be careful here as not all critical damage is visible and we know that big doesn't always equate to strong).

 

Clearly there is limited excess material here by design.  The 'armour effect' (my term) of redundant material is largely gone.  We as an industry will need to adapt processes, procedures and reporting methods to safely use parts that have traditionally been seen as 'rough service' and assumed to be tough in all aspects, but now can be much more easily, critically impaired before installation.

 

In terms of in-service threats - Using the context of aerospace operations, as a fellow involved in winter operations for much of my career, the first thing I look at when I see a part like this (or the 777 folding wing tip) is, how will this part withstand water and de-icing fluid ingress, freezing or, in the case of anti-icing fluids, high pressure rehydration ?  While weight savings comes from material reduction, that means more holes or porous structures,.  Some will be benign, others may introduce new vulnerabilities or stress points when a contaminant freezes or expands in rehydration.   

 

So a thought for the future - How does an engineer using this new capability train the generator to take a defensive approach with regard to handling or in-service abuse?

 

Thanks again for expanding the knowledge base here.  Always a worthy read.

 

John

 

 

[cloakfiend 996]

@SteveCox3D Nice quality prints! great polish and cleanup too!! I wish I could cast metal.

 

I've been looking into investment casting, but small scale set ups are difficult to get decent quality from unless you are doing simple shapes. Centrifugal machines do deal with this to some extent but not ideal for all shapes and sizes. Also having a furnace at home for burning out the PLA is a bit of an issue.....Still working on alternative metal option.... in the meantime I'm streamlining my plating process using 3D printed parts to hold all the consumables and create a modular setup that can easily be up or downscaled accordingly and cheaply!!!! 

 

Great read. Must investigate this 

[SteveCox3D 87]

2 hours ago, JohnInOttawa said:

That's a great article Steve.  Thank you.

 

The benefits of the weight savings are incredible and the part looks nothing like I would imagine, so under the heading of 'thinking outside the box', this approach is game changing.  Not only is there savings in building the part itself, but there is huge potential to save fuel, reduce emissions and, conversely, make parts much, much stronger than they are today for an equivalent weight.   And guys like me are very partial to strong parts ?

 

What follows is largely beyond the scope of the original discussion, but I see it as an inevitable crossroads as we refine these kinds or parts.

 

One of the wild cards we seem to struggle with still is how to handle abuse cases in storage, handling, installation and in-service damage. 

 

We have an industry that grew up on over-built, over-engineered parts.  An unintended benefit of some (not all) of these components is that they could withstand (often undocumented) conditions outside of the original design scope without exhibiting damage.   (I need to be careful here as not all critical damage is visible and we know that big doesn't always equate to strong).

 

Clearly there is limited excess material here by design.  The 'armour effect' (my term) of redundant material is largely gone.  We as an industry will need to adapt processes, procedures and reporting methods to safely use parts that have traditionally been seen as 'rough service' and assumed to be tough in all aspects, but now can be much more easily, critically impaired before installation.

 

In terms of in-service threats - Using the context of aerospace operations, as a fellow involved in winter operations for much of my career, the first thing I look at when I see a part like this (or the 777 folding wing tip) is, how will this part withstand water and de-icing fluid ingress, freezing or, in the case of anti-icing fluids, high pressure rehydration ?  While weight savings comes from material reduction, that means more holes or porous structures,.  Some will be benign, others may introduce new vulnerabilities or stress points when a contaminant freezes or expands in rehydration.   

 

So a thought for the future - How does an engineer using this new capability train the generator to take a defensive approach with regard to handling or in-service abuse?

 

Thanks again for expanding the knowledge base here.  Always a worthy read.

 

John

 

 

Many thanks @JohnInOttawa for contributing to this discussion and the points you make are absolutely valid to this new way of designing.

 

Firstly Generative Design creates surfaces that we would not ordinarily design ourselves, but when those surfaces are derived from the results of stress analysis it's fascinating to recognise the echoes of the way nature designs.  Maybe nature is the greatest designer of all, because it has evolved to make the most effective use of the materials around it.

 

What you've also outlined is something I'm working on at the moment in terms of aiding people to understand the potential pitfalls of Generative Design.  One of those is deploying Robustness & Reliability Engineering methodologies to ensure that the part can perform not just it's "Ideal Function" but also be resilient to the "Noise Factors" that components and assemblies are subjected to.  These can be things such as the abuse loading, and environmental factors as you mention.

 

I'm about to write a presentation that raises the awareness of this.  One example I intend to use is that Generative Design will design you a very lightweight chair based upon four contact areas to the ground and a specified seating load.  That will work under normal circumstances, but we all have a habit of rocking back sometimes onto two legs of a chair, and doing that that might take the design into an unsafe area because that load case was never considered in the original set up.    

 

We're at the start of our journey with these new design techniques, and there's still lots to learn - but it's going to be an exciting ride!  

Edited October 11, 2018 by SteveCox3D

[SteveCox3D 87]

1 hour ago, cloakfiend said:

@SteveCox3D Nice quality prints! great polish and cleanup too!! I wish I could cast metal.

 

I've been looking into investment casting, but small scale set ups are difficult to get decent quality from unless you are doing simple shapes. Centrifugal machines do deal with this to some extent but not ideal for all shapes and sizes. Also having a furnace at home for burning out the PLA is a bit of an issue.....Still working on alternative metal option.... in the meantime I'm streamlining my plating process using 3D printed parts to hold all the consumables and create a modular setup that can easily be up or downscaled accordingly and cheaply!!!! 

 

Great read. Must investigate this 

@cloakfiend It's interesting to see in this case that the optimised part, which had much thinner sections, had a much better burn out of the PLA than the original baseline part. So it seems like the shapes can be fairly complex, with thin walls and sections, and still create good quality cast parts  

Edited October 11, 2018 by SteveCox3D

[JohnInOttawa 104]

I only wish I had the neurons to process the breath of this discussion.  You're clearly on a solid path to important change.  Thanks for bringing that discussion here.  Inspiring!

 

Now, perhaps you can help me find a less discouraging way to explain to our maintenance and engineering staff how we signed out their airplane in perfectly good shape, but returned it broken....  There really is no good way to say, 'the lavatory flush needs repair' and 'have a good one' in the same sentence.

 

J

 

 

[cloakfiend 996]

I'd like to use that generative stuff for shelving or platforms to hold weight. Or a platform holder with multiple weight bearing sections, so you can vary the positioning whilst maintaining optimal strength. Yet another thing on the list of things to do.....

[geert_2 558]

Thanks for this great article. The first thing that came into my mind was that it looked like animal bones. So, after all, nature did a good job...

 

What I am not sure about is the holes and sharp bends near the 4 mounting holes. I think I would rather close these holes and smoothen the sharp bends out a little bit. But that is just my personal feeling. I am from the stone-age, with Flintstone-like designs. Also, we are in a school here, even though a university, so everything needs to be built in solid concrete to survive.

 

It would be good if you could give feedback on how this design behaves in real life tests? In a real airplane, can it handle high side-loads, which might occur due to turbulence or vortices, especially with flaps deployed? And can it withstand heavy vibrations and hammering of the air (I don't know the correct English term)? See the flap-vortices in the landing plane, and the powerfull blasts in the flying plane.

 

 

 

And in cars, how would such designs survive crash tests? Or could flexibility or deformability be built-in, to absorb energy and save passenger lives?

 

Your "rocking chair", or kids jumping on the bed, are good examples of unforeseen overloads, which such designs should also be calculated for.

 

Anyway, it seems a promising concept with a lot of potential. Looking out for your next posts in this area.

 

A bit off-topic: concerning weight saving in transport: this is why I do not believe in electrical cars, at least not in the near future. Designers do their very best to save a few 100kg of structural weight to maximise efficiency. And then they add a 1000kg battery... which allows to drive only 200km in real life circumstances (=at high speed, and in start-stop traffic jams, with airco/heating on, with radio on, lights on). And then "refueling" takes a whole day. While with a diesel engined car I can drive 1200km, and refueling takes 3 minutes. The average weight balast of the diesel fuel on such a trip is 25kg (=max weight / 2). Then there is the problem that an electric car consumes as much energy as 100 house-holds. Here in Belgium, the light is already likely to go out this winter, due to shortage of electricity. A diesel has an efficiency of ca. 40%; while an electric car has an efficiency between 5% and 10%, if you include losses in electricity production, transport, batterycharger, battery itself, converter, motor. I sometimes have the impression that electric cars are mainly promoted by people who have no clue about technology and who can't calculate, or people who have a dark agenda... But that is a different topic...

 

[JohnInOttawa 104]

You've raised a number of different issues here.  I can only partially address the airborne ones.

 

Part fatigue and resistance to aerodynamic loading and shock (I will share your disclaimer on terminology) is sometimes actually worsened by excessive material.  It's a funny paradox, but here is an example.

 

We had a small aircraft manufacturer whose design developed an in-service issue with flap asymmetry.  So, flaps extended to the selected value on one side, but only partially to that extent on the other.  This introduced roll forces that, while within the normal roll control capability of the aircraft,  presented an unacceptable level of threat.  The initial investigation pointed to an actuator assembly that appeared to be failing due to unexpected shock loading when taxying with flaps extended, over a rough surface.  The load path up the gear and through the gear to wing attach point amplified the forces experienced when hitting a bump.

 

The manufacturer elected to address the problem by beefing up the actuator assembly.  The service performance worsened!  Adding new material inadvertently reinforced a previously unknown load path directly to the failure point.  The entirety of the issue was, however revealed, as the new version's material deformation along the load path pointed to the issue.  The fix was to go with less material, but designed to ensure the loads followed the expected load path. 

 

There's also a saying we often use with passengers, especially when we are travelling in uniform, in the cabin, during turbulence.  One of the most frequent questions concerns all of the wing flexing going on out the window.  That which flexes as designed doesn't break and doesn't simply pass all of that load into the cabin as a stiffer implementation might.  Again, more material tends to mean more stiffness, so fighting forces rather than directing them.

 

I have been amazed at some of the organic designs showing up these days, where a part that is intended to provide rigidity in one plane also has significant flex in another axis, specifically to provide resillience against off-axis jolts, or others where metal constantly flexes without risk of fatigue effects throughout its design life.  I have a harmonic drive sitting on my desk, all steel, but the spline flexes like it could be made of TPU.  I am told that this is designed below the fatigue threshold of the metal (Steve, you will have the right term for it), so the steel can flex like this 24/7 for the rest of my life with no issues.

 

Sorry for the long post, it reflects just how fascinating this subject is, how important these concerns are to an end user like me.

 

BTW, nice shots of the vortices.  One of the reasons we take delays on departure is for wake separation, minimum times and distances depending on the aircraft mix.  This separation is a primary reason for long taxi to takeoff at major hubs during peak periods.  Something to pass the time thinking about next time you're number 20 in line at Heathrow, Toronto or Atlanta.

 

If you ever have a chance  watch an A380 punch through cloud on departure, you might want to compare.  That aircraft can shake traffic 20 miles behind it..

 

Great thread!

John

Edited October 13, 2018 by JohnInOttawa

[laverda 0]

Interesting thread and design. I am not a mechanical engineer although I have designed and sold a few products. I have to question the design at the pivot end though. It seems like the contact area to the pivot is much smaller and both longitudinally and around the circumference. It would have very little side to side strength to keep things aligned properly and I expect it to fail at the pivot. A side impact would easily cause an alignment issue if not a failure. I would think a further spaced contact area around the circumference and triangulation would increase the overall performance considerably without adding much material.

[SteveCox3D 87]

The above contributions to this thread by @geert_2 @laverda and @JohnInOttawa all make great points, especially in relation to the load cases that might be seen by this part either in normal, or even abnormal, use.

 

The part in this case study was produced from a Generative Design set-up using three different load cases acting on the pivot of the part.  None of those three load cases were applying a lateral load to the pivot because it was considered that the connecting part is constrained in such a way that it cannot transmit any lateral load, so the design synthesis has not had to take that into account.

 

One of the key things to realise about Generative Design is that it changes the role of the engineer/designer at the very start of the process. We are used to getting quickly into designing solutions for the part we need, but GD makes us think much more about the problem we are trying to solve.  That means spending more time at the beginning thinking through all of the possible loads, and directions of those loads, that the part will see during it's lifecycle.

 

That in itself can be a challenge because sometimes we don't know all of that information, which is why we tend to over-engineer parts, as John pointed out. To address that I've seen Generative Design being used in conjunction with load data that's been generated from real time, physical testing.  For instance instrumenting a car suspension set-up and then testing it to the absolute limits on a test track will give much more load data that can be input into the process.  That should generate a more robust solution.

 

GD is something that will make us think in a different way, and more deeply, about the things we design in order to fully take advantage of it's opportunities.  It's still very new and it's great to have discussions like this as we start to explore the way that we use it.  

Edited October 15, 2018 by SteveCox3D

[geert_2 558]

Thanks for the further explanation. I start to see the potential value for a lot of designs.

 

I am not an engineer, so I don't have to design structural things where high loads are a concern. But I do like the concept, and understanding of the philosophy can also improve other (simple) designs which need fixings like screws etc. Where we now would use the traditional chunky designs based on machining and injection moulding limitations.

 

Do you use safety margins in this design concept? Like in traditional engineering where you would use a factor two (or whatever) in strength calculations?

 

Would this concept also work for structural designs, like buildings and bridges, which are made out of lots of smaller parts? I wonder what would come out for bridges like the one below? (Edit: would it result in an organic shape or a "technical" shape?) We have a lot of these here in the port of Antwerp. This is a small one, but some others are more than 70m long, to let huge container ships pass through the docks. They move up in a complex way, to keep the counterweight balanced all the time (=the big block at the back).

(Photo from Wikipedia, by user Arafi, CC-BY-SA license)
 

Edit: these bridges not only have to carry the load of trucks, bussen and trains, but also a high wind-load when in the vertical position.

 

 

PS: the films (GIF-animations) of airplanes above are not mine. They were generously provided by Mr. Google. So all credits go to the original photographer. There are lots of other magnificient turbulence photos: search for: "landing airplane vortex" or similar.

 

 

Edited October 16, 2018 by geert_2
Clarified a few things

[laverda 0]

8 hours ago, SteveCox3D said:

The above contributions to this thread by @geert_2 @laverda and @JohnInOttawa all make great points, especially in relation to the load cases that might be seen by this part either in normal, or even abnormal, use.

 

The part in this case study was produced from a Generative Design set-up using three different load cases acting on the pivot of the part.  None of those three load cases were applying a lateral load to the pivot because it was considered that the connecting part is constrained in such a way that it cannot transmit any lateral load, so the design synthesis has not had to take that into account.

 

One of the key things to realise about Generative Design is that it changes the role of the engineer/designer at the very start of the process. We are used to getting quickly into designing solutions for the part we need, but GD makes us think much more about the problem we are trying to solve.  That means spending more time at the beginning thinking through all of the possible loads, and directions of those loads, that the part will see during it's lifecycle.

 

That in itself can be a challenge because sometimes we don't know all of that information, which is why we tend to over-engineer parts, as John pointed out. To address that I've seen Generative Design being used in conjunction with load data that's been generated from real time, physical testing.  For instance instrumenting a car suspension set-up and then testing it to the absolute limits on a test track will give much more load data that can be input into the process.  That should generate a more robust solution.

 

GD is something that will make us think in a different way, and more deeply, about the things we design in order to fully take advantage of it's opportunities.  It's still very new and it's great to have discussions like this as we start to explore the way that we use it.  

7

Steve

Thank you for the clarification. It looks very promising. It would be interesting to see the difference if lateral loads were included. 

Steve

[geert_2 558]

This fascinating organic modeling style now slowly seems to become more mainstream. In a recent robot wars series (Battle Bots 2019, USA) there was a robot Quantum in this style (at left, after it won and threw the other bot out of the arena). Its beak can crush with several tons of force. This is a low-resolution screendump, but you can still see the beauty. These fighter robots are often brilliant pieces of engineering.

 

 

 

[cloakfiend 996]

Lol i used to watch robot wars back in the day. This robot reminds me of razer. My fav bot. They just punched holes in it lol. No automated generative design back in those days!! Lol.